Horton

Re: Fwd: Problem with a "complex" upsert

Dean Rasheed <dean.a.rasheed@gmail.com>

View thread View on postgr.es
From: Dean Rasheed <dean.a.rasheed@gmail.com>
To: Amit Langote <Langote_Amit_f8@lab.ntt.co.jp>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, Mario De Frutos Dieguez <mariodefrutos@gmail.com>, pgsql-bugs@lists.postgresql.org
Date: 2018-08-03T09:40:50Z
Lists: pgsql-bugs

Attachments

On 3 August 2018 at 07:52, Dean Rasheed <dean.a.rasheed@gmail.com> wrote:
> This doesn't look right to me. It breaks the following case ...

Here's an updated patch that fixes this.

> I also don't see why it should reject columns from the view that
> aren't in the base relation.

This patch also allows access to view columns that aren't in the
underlying base relation. The rationale for the result in the new test
case where it attempts to insert (1,'y') into columns (aa,bb) of the
view is that the new view row that would have resulted if the insert
had succeeded is ('y',1,(1,'y')), hence that's what excluded.* should
be for the view in the "on conflict" action, and there should be no
problem referring to any part of that excluded view tuple.

Regards,
Dean

Commits

  1. Fix INSERT ON CONFLICT UPDATE through a view that isn't just SELECT *.